Why Robolitics™ is a Unique Solution
Robolitics™ is real-time Monitoring, Alerting & Reporting platform providing cross Asset Class and cross Regulation surveillance, integrating them into a single platform.
It uses Machine Learning and AI to reduce the number of false positives and identify new risks and issues as they happen, rather than after the event. It enables you to group issues and deal with alerts faster by clustering, using powerful visualisation tools and providing a detailed historic review of previous events. Robolitics™ enables you to reduce costs whilst increasing the quality and completeness of your surveillance.
There are four primary drivers behind the creation of Robolitics™
- The technology platforms the current market leaders are built on are reaching the end of their technology life cycle
- The regulatory changes demand deeper analytics across broader data sets, and the expectation is that this should be in real time rather than T+1
- Regulators are beginning to demand automation. Increased automation and machine-supported learning will become critical tools for cutting costs in what is, at present, primarily a manual alert management process. Manual approaches are becoming prohibitively expensive for financial institutions.
- Current solutions focus on individual regulations, or are mash-ups of different products. Non-integrated products under a common brand are extremely costly and complex, and do not provide cross business compliance solutions.
AI and Machine Supported Learning
Robolitics™ does not use cascading filters and rule sets. Instead, we use four Machine Learning principles to identify, categorise, and prioritise an alert:
- Risk based categorisation
- Inference Clustering
- Deep Learning and Commonality
Intersection - enables the business to interact with the data, looking for intersecting conditions on alerts or multiple alert categories. This approach supports compliance in understanding the drivers of the alert and how these are changing over time.
Risk Based Categorisation - We also enable users to manage the alert sensitivity by client risk category dialling down sensitivity for low-risk clients and dialling up for high0risk clients. We create clusters to simplify and reduce the effort of analysis.
Inference Clustering – Outcome based learning Robolitics™ uses text-based analysis of the alert workflow to create dynamic categories and build clusters based around these categories.
Deep Learning and Commonality – Deep learning is about identifying broader patterns across all data. Robolitics™ is able to identify patterns in the data, e.g. similar streets or post codes, rather than identical addresses or transaction patterns linked to specific accounts or sudden changes in payment behaviour by a number of clients.
Deep learning is particularly strong at identifying new categories of behaviour or risk
Why Current Platforms Cannot Meet the Demands of the Regulators
The platforms sold by the current market leaders use a rule-based approach to dial out the probable false positives.
The problem with rule-based models is that they are time consuming to build, and they very quickly become unwieldy and cumbersome. Eventually, there comes a time when nobody really knows how the rules work or how many real issues are being filtered out.
In addition, rule-based models are:
- Slow to evolve, difficult to tune, & complex to adjust – requiring cycle after cycle of testing;
- Poor at responding to change’
- Typically additive only, and seldom simplified: they frequently contain repeating or nested rule sets;
- At higher risk of eliminating real events as part of the process of reducing False Positives;
- Likely to end up with an embedded history, which becomes a fossil record of the threats and risks faced by the business over the years.
The Robolitics™ approach is different. We use the latest technology to improve the identification of real events and allow the compliance teams to interact using visual tools to understand what is going on rather than react to the avalanche of false alerts. This greater understanding enables the team to take a more strategic approach to managing real risks and issues, setting and changing the risk appetite of the business as the market changes and evolves.
Market Leading KYC
“The KYC process is a simple …. but its seldom done well”.
The challenge is not the initial on-boarding process but the ongoing problems of:
- Being able to identify and retrieve documentation when needed to prove the process (Heliocor has DocStore a Block Chain Repository for Doc Management compliant with GDPR)
- Managing the re-use of documentation if people are linked to multiple accounts or legal entities (Heliocor uses Personas to Control)
- Managing client risk and product approvals & Keeping up to date re-KYC’ing, increasingly important under MiFID (Using both KYP & Risk Alerting).
DocStore - Secure Compliant Updating KYC Documentation
A secure mobile application that can be integrated into your own Mobile banking app, a secure Repository for documents exploiting Block Chain, and integration into the Robolitics platform for managing alerting and updating.
Mobile app module features include
- Document Management Core and extendable doc types
- Document management
- Machine readability & document details
- A secure store for documents, which holds key documents and manages updates so that the newest is always available, minimising KYC update process. Stores document metadata and machine reads data from the document.
KYC process integration
- KYC process and integration with on-boarding
- Automation of the KYC document management and storage process
- Supports EDD process
- Persona management
- Federation of documentation between personas to reduce Document update workload and mobile app and alerting internally and via client mobile app component
Load Management & Creation of New Analytic Robots
Peak load management - In addition to the four methods described previously, Robolitics can be configured to cope with sudden high volume events such as Brexit Vote Decision or Donald Trump’s election as President. During such periods, the volume of alerts are managed back.
However, because Robolitics™ stores all event (both above and below the radar), when market conditions demand it, the alert parameters can be narrowed for individual or all alerts, keeping volumes within control and enabling un-investigated items to be reviewed later as a cluster or surfaced as part of any future review of historic behaviour.
Creating New Alerts – full license of Robolitics™ includes access to the RDK, which enables new alerts to be created quickly. Such alerts can be generated to spot specific types of behaviour for internal control and management as well as ensuring conformance to regulations.